Logo

IARA - Interpretable AI for Risk Assessment

IARA is a next-generation in silico New Approach Methodology (NAM) that combines artificial intelligence with a curated human Biomedical Knowledge Graph (BKG) to predict liver, cardiac, nervous system and kidney toxicity.

📈 Model Performance

OrganCV ROC-AUCTest ROC-AUCAccuracyMCCSensitivitySpecificityF1 ScoreYouden threshold (J)
HEART0.74 ± 0.020.79 ± 0.020.71 ± 0.060.44 ± 0.100.69 ± 0.130.75 ± 0.110.73 ± 0.120.75 ± 0.09
LIVER0.87 ± 0.010.84 ± 0.010.72 ± 0.050.46 ± 0.060.78 ± 0.080.68 ± 0.120.69 ± 0.030.35 ± 0.11
NERVOUS SYSTEM0.82 ± 0.010.82 ± 0.010.73 ± 0.030.48 ± 0.050.78 ± 0.080.69 ± 0.090.73 ± 0.030.47 ± 0.10
KIDNEY0.77 ± 0.010.64 ± 0.020.62 ± 0.030.23 ± 0.060.65 ± 0.070.58 ± 0.080.65 ± 0.040.55 ± 0.09

Select one of the two available query modes to perform toxicity prediction. Choose between a compound identifier–based workflow or a protein target–based workflow.

COMPOUND IDENTIFIER-BASED QUERIES

Submit a chemical identifier (e.g., ChEMBL ID, InChlKey, CID) of the query compound.

CHEMBL ID
Name
CHEMBL418971
Bisphenol A
CHEMBL1076347
Triclocarban
CHEMBL300764
(S)-3-Acetylamino-2,2-difluoro-4-phenyl-butyric acid methyl ester
CHEMBL1631217
Hydroxyitraconazole
CHEMBL1467
Allopurinol

PROTEIN TARGET-BASED QUERIES

Submit a list of proteins associated with the query compound, optionally including weight score (0 - 10) for each protein to indicate the strenght of confidence of each compound-protein interaction.

Uniprot ID
Weight
P03372
7
P10275
6
Q14994
7